Características de la red social de PPK

respectively.

Genotyping, population structure, kinship and linkage disequilibrium

Genomic DNA was extracted from 14-day-old plants accord- ing to Stein et al. (2001). The accessions were genotyped on the Illumina iSelect 50 k Barley SNP Chip (Illumina) at Trait Genetics GmbH (Gatersleben, Germany). Physical positions of markers were taken from Bayer et al. (2017), which is based on the barley pseudo-molecule assembly by Mascher et al. (2017). SNPs having failure rates > 10%, heterozygous calls > 12.5% and a minor allele frequency (MAF) < 5% were excluded from the analyses, as well as unmapped SNPs. Thus, 33,818 SNPs were left for subse- quent GWAS. In order to calculate the kinship matrix and the population structure, the markers were further filtered with the software PLINK 1.9 (www.cog-genom ics.org/plink /1.9/) (Chang et al. 2015). The tool LD prune was used with the following parameters: indep pairwise window size 50, step 5 and an r2 _{threshold 0.5 (Campoy et al. 2016). This resulted}

in 8533 markers for calculating the kinship and population structure. Kinship was calculated with the web-based plat- form Galaxy (Afgan et al. 2016) using the tool Kinship and the modified Roger’s distance (Reif et al. 2005). Population structure was determined with the software STRU CTU RE v2.3.4 (Pritchard et al. 2000). In order to identify the optimal subpopulations, an admixture model was used with a burn-in of 50,000, followed by 50,000 Monte Carlo Markov chain (MCMC) replications for k = 1 to k = 10 with 10 iterations. STRU CTU RE HARVESTER (Earl and vonHoldt 2012) was used to identify the optimal k. Following this, a new STRU CTU RE analysis was performed with a burn-in of 100,000 and 100,000 MCMC iterations at the optimal k value. Acces- sions were considered as admixed, when their membership probabilities were < 80% (Richards et al. 2017). Linkage disequilibrium (LD) was calculated as squared allele fre- quency correlations (R2_{) between all intra-chromosomal}

marker pairs using the tool linkage disequilibrium in the web-based platform Galaxy. Genome-wide LD decay was plotted as R2 _{of a marker against the corresponding genetic}

distance, and a Loess regression was computed. For R2_{, the}

default settings were used (Sannemann et al. 2015).

Genome‑wide association studies

Genome-wide association studies (GWAS) were performed using the Galaxy implemented tool GAPIT, which uses the

R package GAPIT (Lipka et al. 2012). The model used was

a compressed mixed linear model (CMLM)(Zhang et al.

2010) including the population structure (Q) and kinship (K). In order to detect significant marker–trait associations,

a Bonferroni correction was employed. For this, the reduced marker set of 8533 markers, which was used for calculating population structure and kinship, and a significance level of

P = 0.2 was used (Muqaddasi et al. 2017; Storey and Tibshi- rani 2003). This resulted in a threshold of − log₁₀ (P) ≥ 4.63. GWAS for greenhouse trials, and field trials in Australia were conducted for each isolate separately. For field trials in Zhodino and Quedlinburg, GWAS were conducted across years for each location. Manhattan plots were generated with the R v.3.4.4 package qqman.

In order to compare previously described QTL with QTL identified in the present study, the databases GrainGenes (https ://wheat .pw.usda.gov/GG3/) and BARLEX (https :// apex.ipk-gater slebe n.de/apex/f?p=284:10) were used to obtain marker information and identify physical positions of previously published QTL studies. Where previously described QTL were identified based on iSelect markers, the physical positions were obtained from Bayer et al. (2017).

Results

Phenotypic evaluation of greenhouse trials

Phenotyping in the greenhouse with three different isolates showed a wide range of variability in the infection response type (1–10 scale) for all three isolates tested. The average infection response type (IRT) for isolate Hoehnstedt ranged from 1 to 9 (mean 3.96), for No 13 from 1 to 10 (mean 5.28) and for NFNB 50 from 1 to 10 (mean 3.6) (Fig. 1). Analysis of variance showed significant differences among the barley accessions for seedling resistance to NFNB for all isolates (Table 1). In trials with isolates No 13 and Hoehnstedt, the proposed differential set by Afanasenko et al. (2009) was used as a reference. The infection scores for the differential lines can be seen in Table 2. For isolate No 13, the infection scores ranged from 0.75 (CI 5791) to 8 (Harrington). For

Hoehnstedt, the lines showed less variance and the scores

ranged from 2.63 (Harbin) to 4.89 (CLS 25282). For isolate

NFNB 50, the lines used as references and their respective

infection scores can be obtained from Table 3. The scores ranged from 0.8 (Beecher) to 9.1 (Grimmett).

Phenotypic evaluation of field trials

A wide range of disease severity was observed for all three locations (Fig. 2; Table 1).

In Germany in both years, infection pressure for NFNB was high in Summer Hill trials. Disease severity scores ranged on average between 4.1 and 31.5% (mean 11.5%). The frequency distribution was slightly right skewed with 184 accessions showing disease severity of < 10% and 7

accessions showing scores of > 25%. The heritability for this location was estimated at h2 _{= 0.73.}

In Belarus in 2017, conditions were unfavourable for NFNB yet favourable for powdery mildew, which did not allow any more than two assessments of net blotch. Hence, for this location the disease score based on the respective last scoring date was used to calculate the mean disease severity across years. Disease severity scores ranged on

average between 0.1 and 60% (mean 7.7%). The frequency distribution for this location is right-skewed with 201 accessions showing a disease severity of < 5% and 5 acces- sions showing scores of 40% and higher. Heritability for

Fig. 1 Frequency distribution for Pyrenophora teres f. teres reaction after inoculation with isolates Hoehnstedt, No 13 and NFNB 50

Table 1 Analysis of variance (ANOVA) for net form of net blotch (NFNB) severity for 449 barley genotypes evaluated under green- house and field conditions

Isolate/location Effect F value P value

Hoehnstedt Genotype 4.66 < 0.0001 No 13 Genotype 19.12 < 0.0001 NFNB 50 (greenhouse) Genotype 22.15 < 0.0001 Quedlinburg Genotype 3.68 < 0.0001 Zhodino Genotype 4.63 < 0.0001 NFNB 50 (field) Genotype 11.9 < 0.0001 NFNB 73 Genotype 12.29 < 0.0001 NFNB 85 Genotype 12.61 < 0.0001

Table 2 Disease severities of differential lines (Afanasenko et  al.

2009) used in field trials in Belarus and Germany, and in greenhouse trials with isolates No 13 and Hoehnstedt

a _{% of leaf area infected}

b _{infection response type based on Tekauz (1985), 1 to 9 scale}

Differential line Trial

Belarusa _Germanya _{No 13}b _Hoehnstedtb

Harrington 19.75 12.20 8.00 4.75 Skiff 13.5 7.49 6.13 3.22 Prior 3 5.52 7.63 – CI 9825 1.5 10.27 1.56 2.44 Harbin 3 11.79 1.38 2.63 K 20019 3 10.19 1.63 2.89 CI 5791 0.75 14.51 0.75 2.86 CLS 25282 0.75 14.11 1.17 4.89 K 8755 1 14.06 1.88 3.38

2639 Theoretical and Applied Genetics (2019) 132:2633–2650

1 3